This tutorial uses an open source Python package named AI Fairness 360 or AIF360.
Additionally, download datasets following instructions at https://github.com/IBM/AIF360/tree/master/aif360/data
Alternatively, instructions are provided below for manually installing AIF360 and downloading datasets using Conda on Windows
Create and activate environment
conda create --name aif360 python=3.6
conda activate aif360Clone AIF360 from GitHub:
git clone https://github.com/IBM/AIF360Install R-essentials for downloading MEPS data
conda install -c r r-essentialsDownload datasets and place under appropriate folders under AIF360/aif360/data/raw by cloning this repository (NOTE: clone at same level as AIF360) and running the belowmentioned notebooks in the root folder
git clone https://github.com/monindersingh/pydata2018_fairAI_models_tutorial.gitChange to the root folder of just cloned repository and run
jupyter notebook pydata_datasets.ipynb
jupyter notebook pydata_meps_datasets.ipynbThen, navigate to the root directory of the cloned AIF360 project and run:
pip install .Finally, install the additional requirements as follows:
conda install ecos
pip install -r requirements.txt